基于迁移学习的水产动物图像识别方法

doi:10.11772/j.issn.1001-9081.2017102487

计算机应用 ›› 2018, Vol. 38 ›› Issue (5): 1304-1308.DOI: 10.11772/j.issn.1001-9081.2017102487

基于迁移学习的水产动物图像识别方法

王柯力, 袁红春

上海海洋大学信息学院, 上海 201306

收稿日期:2017-10-20 修回日期:2017-11-29 出版日期:2018-05-10 发布日期:2018-05-24
通讯作者: 袁红春
作者简介:王柯力(1990-),男,湖南益阳人,硕士研究生,主要研究方向:深度学习、图像识别、数据挖掘、人工智能;袁红春(1971-),男,江苏海门人,教授,博士,主要研究方向:智能计算、智能信息处理。
基金资助:
国家自然科学基金资助项目（41776142）；上海市科学技术委员会支撑项目（1439190400）。

Aquatic animal image classification method based on transfer learning

WANG Keli, YUAN Hongchun

College of Information Technology, Shanghai Ocean University, Shanghai 201306, China

Received:2017-10-20 Revised:2017-11-29 Online:2018-05-10 Published:2018-05-24
Contact: 袁红春
Supported by:
This work is partially supported by the National Natural Science Foundation of China (41776142), the Support Project of Shanghai Science and Technology Commission (1439190400).

摘要/Abstract

摘要： 针对传统水产动物图像识别方法步骤复杂、准确率差、泛化性差，而深度卷积神经网络（DCNN）模型开发难度大等问题，提出一种基于参数迁移策略采用微调方式再训练源模型的方法。首先，对图像进行数据增强等预处理；然后，在修改源模型全连接分类层的基础上，进一步将高层卷积模块的权重设置为可训练以进行自适应调整；最后，以验证集识别率与训练时间作为评估指标，针对不同源模型采用不同的可训练参数占比进行性能实验。实验结果表明，通过再训练得到的图像识别模型准确率可达到97.4%，相比源模型最多可提高20个百分点；在可训练参数占比为75%左右时可得到较理想的性能。通过实验证实了采用微调方法可以在低成本开发条件下得到性能良好的深度神经网络图像识别模型。

关键词: 水产动物图像, 深度卷积神经网络, 迁移学习, 微调, 数据提升

Abstract: Aiming at the problems that traditional aquatic animal image recognition methods have complex steps, low accuracy and poor generalization, and it is difficult to develop Deep Convolutional Neural Network (DCNN) model, a method based on parameter transfer strategy using fine-tune to retrain pre-trained model was proposed. Firstly, the image was preprocessed by data enhancement and so on. Secondly, on the basis of modifying the source model's fully connected classification layer, the weights of high-level convolution modules were set to be trained for adaptive adjustment. Finally, using training time and recognition accuracy on validation set as the evaluation indexes, the performance experiments were conducted on various network structures and different proportion of trainable parameters. The experimental results show that the highest retrained model classification accuracy can reach 97.4%, 20 percentage points higher than the source model, the ideal performance can be obtained when the proportion of trainable parameters is around 75%. It is proved that the fine-tune method can obtain a deep neural network image classification model with good performance under low-cost development condition.

Key words: aquatic animal image, Deep Convolutional Neural Network (DCNN), transfer learning, fine-tune, data enhancement

中图分类号:

TP391.4

王柯力, 袁红春. 基于迁移学习的水产动物图像识别方法[J]. 计算机应用, 2018, 38(5): 1304-1308.

WANG Keli, YUAN Hongchun. Aquatic animal image classification method based on transfer learning[J]. Journal of Computer Applications, 2018, 38(5): 1304-1308.

参考文献

[1] 朱从容. 计算机视觉技术在水产养殖中的应用[J]. 浙江海洋学院学报:自然科学版, 2008, 27(4):439-443.(ZHU C R. Application of computer vision technology in aquaculture[J]. Journal of Zhejiang Ocean University (Natural Science), 2008, 27(4):439-443.)
[2] NAIBERG A, PETRELL R J, SAVAGE C R, et al. Stereo video technique to size fish in sea cage and tanks[J]. Aquaculture Engineering, 1993, 16(1):393-402.
[3] GAO J, FAN W, JIANG J, et al. Knowledge transfer via multiple model local structure mapping[C]//KDD 2008:Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York:ACM, 2008:283-291.
[4] ZHANG Y, YEUNG D Y. Transfer metric learning by learning task relationships[C]//KDD 2010:Proceedings of the 16th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York:ACM, 2010:1199-1208.
[5] COOTES T F, TAYLOR C J. Data driven refinement of active shape model search[C]//Proceedings of 7th British Machine Vision Conference. Edinburgh, UK:University of Edinburgh, 1996:383-392.
[6] MAHMOOD A, KHAN S. Correlation-coefficient based fast template matching through partial elimination[J]. IEEE Transactions on Image Processing, 2012, 21(4):2099-2108.
[7] VIOLA P, JONES M. Rapid object detection using a boosted cascade of simple features[C]//CVPR 2001:Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington, DC:IEEE Computer Society, 2001:511-518.
[8] SCHMIDHUBER J. Deep learning in neural networks:an overview[J]. Neural Networks, 2015, 61:85-117.
[9] LI S, KWOK J T, ZHU H, et al. Texture classification using the support vector machines[J]. Pattern Recognition, 2003, 36(12):2883-2893.
[10] 丁蓬莉, 李清勇, 张振, 等. 糖尿病性视网膜图像的深度神经网络分类方法[J]. 计算机应用, 2017, 37(3):699-704. (DING P L, LI Q Y, ZHANG Z, et al. Diabetic retinal image classification method based on deep neural network[J]. Journal of Computer Applications, 2017, 37(3):699-704.)
[11] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL].[2017-05-10]. https://arxiv.org/abs/1409.1556.
[12] SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2016:2818-2826.
[13] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2016:770.
[14] 余化鹏, 张朋, 朱进. 基于深度迁移学习的人脸识别方法研究[J]. 成都大学学报(自然科学版), 2017, 36(2):151-156. (YU H P, ZHANG P, ZHU J. Study on face recognition method based on deep transfer learning[J]. Journal of Chengdu University (Natural Science Edition), 2017, 36(2):151-156.)
[15] RUSU A A, RABINOWITZ N C, DESJARDINS G, et al. Progressive neural networks[EB/OL].[2017-05-10]. https://arxiv.org/abs/1606.04671.
[16] LONG M, CAO Y, WANG J, et al. Learning transferable features with deep adaptation networks[EB/OL].[2017-05-10]. https://arxiv.org/abs/1502.02791.
[17] DAI W Y, YANG Q, XUE G R, et a1. Boosting for transfer learning[C]//Proceedings of the 24th International Conference on Machine Learning. New York:ACM, 2007:193-200.
[18] PAN S J, NI X, SUN J T, et al. Cross-domain sentiment classification via spectral feature alignment[C]//WWW 2010:Proceedings of the 19th International Conference on World Wide Web. New York:ACM, 2010:751-760.
[19] TOMMASI T, ORABONA F, CAPUTO B. Learning categories from few examples with multi model knowledge transfer[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(5):928-941.
[20] SZEGEDY C, IOFFE S, VANHOUCKE V, et al. Inception-v4, Inception-ResNet and the impact of residual connections on learning[EB/OL].[2017-05-10]. https://arxiv.org/abs/1602.07261.

基于迁移学习的水产动物图像识别方法

Aquatic animal image classification method based on transfer learning

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	陈争涛, 黄灿, 杨波, 赵立, 廖勇. 基于迁移学习的并行卷积神经网络牦牛脸识别算法[J]. 计算机应用, 2021, 41(5): 1332-1336.
[2]	王锦凯, 贾旭. 基于迁移孪生非负矩阵分解的静脉识别算法[J]. 计算机应用, 2021, 41(3): 898-903.
[3]	刘晓龙, 王士同. 渐进式分离的开放集模糊域自适应算法[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3127-3131.
[4]	曹建芳, 闫敏敏, 贾一鸣, 田晓东. 融合迁移学习的Inception-v3模型在古壁画朝代识别中的应用[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3219-3227.
[5]	赵津, 宋文爱, 邰隽, 杨吉江, 王青, 李晓丹, 雷毅, 邱悦. 儿童阻塞性睡眠呼吸暂停计算机人脸辅助诊断综述[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3394-3401.
[6]	黄学雨, 徐浩特, 陶剑文. 具有特征选择的多源自适应分类框架[J]. 计算机应用, 2020, 40(9): 2499-2506.
[7]	张晨曦, 唐曙, 唐珂. 迁移学习下的火箭发动机参数异常检测策略[J]. 计算机应用, 2020, 40(9): 2774-2780.
[8]	白茹意, 郭小英, 贾春花. 基于特征融合的小样本抽象画图像情感预测[J]. 计算机应用, 2020, 40(8): 2207-2213.
[9]	马金林, 魏萌, 马自萍. 基于深度迁移学习的肺结节分割方法[J]. 计算机应用, 2020, 40(7): 2117-2125.
[10]	肖禾, 刘志勤, 王庆凤, 黄俊, 周莹, 刘启榆, 徐卫云. 基于迁移学习的多视角乳腺肿块和钙化簇分类方法[J]. 计算机应用, 2020, 40(5): 1460-1464.
[11]	谌贵辉, 易欣, 李忠兵, 钱济人, 陈伍. 基于改进YOLOv2和迁移学习的管道巡检航拍图像第三方施工目标检测[J]. 计算机应用, 2020, 40(4): 1062-1068.
[12]	陈杰, 张挺, 杜奕. 基于自适应深度迁移学习的多孔介质重构[J]. 计算机应用, 2020, 40(4): 1231-1236.
[13]	边小勇, 费雄君, 穆楠. 基于尺度注意力网络的遥感图像场景分类[J]. 计算机应用, 2020, 40(3): 872-877.
[14]	郑宗生, 胡晨雨, 姜晓轶. 基于改进的最大均值差异算法的深度迁移适配网络[J]. 计算机应用, 2020, 40(11): 3107-3112.
[15]	毛文涛, 杨超, 刘亚敏, 田思雨. 面向轴承早期故障检测的多尺度残差注意力深度领域适配模型[J]. 计算机应用, 2020, 40(10): 2890-2898.